Modeling Stylized Facts in FX Markets with FINGAN-BiLSTM: A Deep Learning Approach to Financial Time Series.

IF 2 3区 物理与天体物理 Q2 PHYSICS, MULTIDISCIPLINARY
Entropy Pub Date : 2025-06-14 DOI:10.3390/e27060635
Dong-Jun Kim, Do-Hyeon Kim, Sun-Yong Choi
{"title":"Modeling Stylized Facts in FX Markets with FINGAN-BiLSTM: A Deep Learning Approach to Financial Time Series.","authors":"Dong-Jun Kim, Do-Hyeon Kim, Sun-Yong Choi","doi":"10.3390/e27060635","DOIUrl":null,"url":null,"abstract":"<p><p>We propose the financial generative adversarial network-bidirectional long short-term memory (FINGAN-BiLSTM) model to accurately reproduce the complex statistical properties and stylized facts, namely, heavy-tailed behavior, volatility clustering, and leverage effects observed in the log returns of the foreign exchange (FX) market. The proposed model integrates a bidirectional LSTM (BiLSTM) into the conventional FINGAN framework so that the generator, discriminator, and predictor networks simultaneously incorporate both past and future information, thereby overcoming the information loss inherent in unidirectional LSTM architectures. Experimental results, assessed using metrics such as the Kolmogorov-Smirnov statistic, demonstrate that FINGAN-BiLSTM effectively mimics the distributional and dynamic patterns of actual FX data. In particular, the model significantly reduces the maximum cumulative distribution discrepancy in assets with high standard deviations and extreme values, such as the Canadian dollar (CAD) and the Mexican Peso (MXN), while precisely replicating dynamic features like volatility clustering and leverage effects, thereby outperforming conventional models. The findings suggest that the proposed deep learning-based forecasting model holds significant promise for practical applications in financial risk assessment, derivative pricing, and portfolio optimization, and they highlight the need for further research to enhance its generalization capabilities through the integration of exogenous economic variables.</p>","PeriodicalId":11694,"journal":{"name":"Entropy","volume":"27 6","pages":""},"PeriodicalIF":2.0000,"publicationDate":"2025-06-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12191802/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Entropy","FirstCategoryId":"101","ListUrlMain":"https://doi.org/10.3390/e27060635","RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"PHYSICS, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0

Abstract

We propose the financial generative adversarial network-bidirectional long short-term memory (FINGAN-BiLSTM) model to accurately reproduce the complex statistical properties and stylized facts, namely, heavy-tailed behavior, volatility clustering, and leverage effects observed in the log returns of the foreign exchange (FX) market. The proposed model integrates a bidirectional LSTM (BiLSTM) into the conventional FINGAN framework so that the generator, discriminator, and predictor networks simultaneously incorporate both past and future information, thereby overcoming the information loss inherent in unidirectional LSTM architectures. Experimental results, assessed using metrics such as the Kolmogorov-Smirnov statistic, demonstrate that FINGAN-BiLSTM effectively mimics the distributional and dynamic patterns of actual FX data. In particular, the model significantly reduces the maximum cumulative distribution discrepancy in assets with high standard deviations and extreme values, such as the Canadian dollar (CAD) and the Mexican Peso (MXN), while precisely replicating dynamic features like volatility clustering and leverage effects, thereby outperforming conventional models. The findings suggest that the proposed deep learning-based forecasting model holds significant promise for practical applications in financial risk assessment, derivative pricing, and portfolio optimization, and they highlight the need for further research to enhance its generalization capabilities through the integration of exogenous economic variables.

用FINGAN-BiLSTM建模外汇市场的风格化事实:一种金融时间序列的深度学习方法。
我们提出金融生成对抗网络-双向长短期记忆(FINGAN-BiLSTM)模型,以准确再现外汇(FX)市场对数回报中观察到的复杂统计特性和风格化事实,即重尾行为、波动聚类和杠杆效应。该模型将双向LSTM (BiLSTM)集成到传统的FINGAN框架中,使生成器、鉴别器和预测器网络同时包含过去和未来的信息,从而克服了单向LSTM架构固有的信息丢失。使用Kolmogorov-Smirnov统计等指标评估的实验结果表明,FINGAN-BiLSTM有效地模拟了实际外汇数据的分布和动态模式。特别是,该模型显著降低了加拿大元(CAD)和墨西哥比索(MXN)等高标准差和极值资产的最大累积分布差异,同时精确地复制了波动聚类和杠杆效应等动态特征,从而优于传统模型。研究结果表明,基于深度学习的预测模型在金融风险评估、衍生品定价和投资组合优化方面具有重要的实际应用前景,并强调需要进一步研究,通过整合外生经济变量来增强其泛化能力。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Entropy
Entropy PHYSICS, MULTIDISCIPLINARY-
CiteScore
4.90
自引率
11.10%
发文量
1580
审稿时长
21.05 days
期刊介绍: Entropy (ISSN 1099-4300), an international and interdisciplinary journal of entropy and information studies, publishes reviews, regular research papers and short notes. Our aim is to encourage scientists to publish as much as possible their theoretical and experimental details. There is no restriction on the length of the papers. If there are computation and the experiment, the details must be provided so that the results can be reproduced.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信